Extending udata

udata is customizable in many ways, just choose yours.


Before trying to code something specific, take a look at all the settings, there may already be some details you can easily customize with a simple setting.


You can totally customize the udata appearence with themes.

See the dedicated section for more details.

NB: A theme is also an entrypoint, but a special one.


Entrypoints are modules or classes loaded by udata to extends its features.

Harvesters (udata.harvesters)

Plugins can expose extra harvesters via the udata.harvesters class entrypoint.

See the Harvesting section for more details

Views (udata.views)

Plugins can expose extra view features via the udata.views module entrypoint including:

  • a blueprint (should be named blueprint)
  • some view filters
  • some hooks implementation


Hooks are small html snippets loaded dynamicaly.

In any template, you can add a placeholder for hooks using one of these syntaxes:

# All snippets concatened
{{ hook('my-custom-hook') }}

# More complex layout with a for loop
{% for widget in hook('my-widgets') %}
<li>{{ widget }}</li>
{% endofr %}

# Optionnal parameters support
{{ hook('my-parametric-hook', arg1, arg2, kw=value) }}

In a view plugin, a hook implementation is a simple decorated function with the context as first argument.

from udata.frontend import template_hook

def my_hook(ctx):  # Will be available in {{ hook('my_hook') }}
    return 'my hook'

@template_hook('another-hook')  # Will be available as {{ hook('another-hook') }}
def my_custom_hook(ctx):
    return 'my custom hook'

A hook can render a template as simplfy as any basic view:

def my_widget(ctx):
    return theme.render_template('my/widget.html, **ctx)

Hooks can be conditionnaly rendered:

@template_hook('conditionnal-hook', when=lambda ctx: 'key' in ctx)
def my_conditionnal_hook(ctx):
    return 'key is present in context and value is {}'.format(ctx['key'])

Hooks can also receive parameters in addition to context:

{{ hook('with-params', arg1, key=value) }}
def my_hook(ctx, arg, **kwargs):
    # Do something with params

Metrics (udata.metrics)

A module entrypoint allowing to register new metrics.

Models (udata.models)

This module entrypoint allows you to expose new models or to extend existing ones by adding new badges or new known extras.

Models entrypoints may also expose migrations in the migrations folder sibling to the models module. If you only need to expose migrations, just provide an empty models module.

This class entrypoint allows to register new link checkers that udata will recognize and use.

Tasks and jobs (udata.tasks)

This module entrypoint allows to register new asynchronous tasks and schedulable jobs.

Previews (udata.preview)

A class entrypoint for preview providers.

These plugins should extend udata.core.dataset.preview.PreviewPlugin.


from udata.core.dataset.preview import PreviewPlugin

class MyPreview(PreviewPlugin):
    def can_preview(self, resource):
        # Check whether or not you can display a preview
        # You can access the resource or its dataset (through resource.dataset)
        # to check your requirements

    def preview_url(self, resource):
        # Return the absolute preview URL for the given resource.
        # You can access the resource or its dataset (through resource.dataset)
        # to build your preview URL

You can mark a preview plugin as fallback, meaning it will only be a candidate if other plugins can’t provide a preview. This is typically for plugin displaying generic preview (ie. only relying on mimetype for example):

from udata.core.dataset.preview import PreviewPlugin

class MyGenericPreview(PreviewPlugin):
    fallback = True

Enabled plugins are cached so don’t forget to flush cache when:

  • you change your PLUGINS configuration
  • you deliver new plugin versions

Generic plugins (udata.plugins)

A module entrypoint for generic plugins. They just have to expose a init_app(app) function and can perform any manual initialization.

Use this entrypoint if you want to perform something not handled by previous entrypoints.

Default settings

Any registered plugin may also expose some default settings in a settings module (ie. my_plugin.settings). They will be automatically discovered and registered.


Any registered plugin may also expose translations in its root module translations directory. They will be automatically discovered and loaded if the plugin is enabled.


Last but not least, if none of the above match your needs, you can also contribute to the core udata project and submit some contributions.

See the Contributing Guide